矿区土壤Cu含量高光谱反演建模

doi:10.3964/j.issn.1000-0593(2018)02-0575-07

Abstract
References
Related Citation (15)

Download: PDF (6210 KB)
Export: BibTeX | EndNote (RIS)

Abstract To explore the feasibility of evaluating soil Cu content with Hyperspectral Remote Sensing method, 83 soil samples were collection from a certain diggings in Hunan Province. Using ASD field spectrometer and Induced Coupled Plasma Atomic Emission Spectrometry collecting the reflectance spectra and Cu content. The reflectance spectra were processing with several method: resampling, first/second derivative, standard normal variate. Based on the transformational spectra, potential modeling variables were selected by using principal component analysis and correlation analysis. Final model with stepwise regression were established. Important wavelengths were recognized that respond to Cu content based on the optimal model. The result showedthat, compared to traditional principal component analysis method, because of retaining the weak spectrum signal, principal component stepwise regression with standard normal variate spectra can improve the accuracy of Cu content estimation (R²=0.86), and the estimation of predicting samples is effective. The residual error of modeling samples and predicting samples is 0.76 and 1.29, and it passed the F test. In study area, the reflectance on 360~400, 922~1 009, 1 833~1 890 and 2 200~2 500 nm was indicative to Cu content. The study result will enrich a typical case of diggings in South of China, and provide theoretical support for developing method of soil environment monitor based on Hyperspectral Remote Sensing.

Key words：Heavy metal; Soil; Hyperspectral remote sensing; Principal component analysis; Stepwise regression

Received: 2017-05-19 Accepted: 2017-10-06

ZTFLH:

TP79

Corresponding Authors: ZOU Bin E-mail: 210010@csu.edu.cn

Cite this article:

URL:

https://www.gpxygpfx.com/EN/10.3964/j.issn.1000-0593(2018)02-0575-07 OR https://www.gpxygpfx.com/EN/Y2018/V38/I02/575